Constellation Energy
Owns/operates ~60 GW of generation (largest US nuclear fleet + Calpine gas/geothermal) and sells the output wholesale into PJM/ERCOT/Northeast, via direct hyperscaler PPAs, and through a competitive retail book; earns on the spread between realized power prices and fuel/purchased-power cost plus capacity/ancillary revenue.
The thesis on this name
State of Data-Center Power
The largest US nuclear fleet operator and the premier counterparty for hyperscaler 20-year nuclear PPAs — Microsoft (Three Mile Island/Crane restart) and Meta (Clinton, from 2027) both signed with Constellation (fact). The Calpine merger adds ~21GW of gas/geothermal, broadening the dispatchable fleet feeding AI load. Nuclear's 24/7 carbon-free baseload is exactly what hyperscalers will pay a premium for, and Constellation owns the scarcest version of it. Trades at ~17-21x EV/EBITDA, a ~50% premium to Vistra (fact) — so the asymmetry is moderate and the entry matters; this is a high-quality demand-anchor bought for the PPA-annuity, sized for a pullback.
State of Data-Center Power
Largest US nuclear baseload + the closed $16.4B Calpine gas fleet = the one vendor that sells hyperscalers 24/7 carbon-free + flexible gas; and it's de-rated ~25% off the high.
State of Data-Center Power
Largest US nuclear fleet + $16.4B Calpine gas; the one 24/7-carbon-free+flexible package for hyperscalers, de-rated ~25% off its high.
State of Nuclear Energy
The lowest-risk way to own the nuclear/AI-power convergence with actual cash flows — included here as durable-compounder optionality (a 'related arm,' reference-weighted). CEG monetizes the data-center power bid via 20-year Meta/Microsoft PPAs and the TMI restart without single-reactor construction or pre-revenue risk. Fwd P/E ~26x and -15.8% YTD as the market recalibrates the premium.
Earnings, margins, COGS & capex
FY2025 revenue was $25.5B (+8.3%) with ~$3.09B operating income (~12% margin) and adjusted operating EPS of $9.39 (fact). The 7 Jan 2026 close of the $16.4B-equity / ~$26.6B-total Calpine acquisition transformed the P&L — Q1 FY26 revenue jumped +64% to $11.1B and management guides FY26 adjusted EPS to $11–12 (affirmed), implying ~20%+ EPS accretion. The economics are a power-price-minus-fuel spread business: nuclear is low-marginal-cost baseload (high-margin), Calpine adds gas with real fuel/purchased-power cost (Calpine fuel/purchased-power ~$6.35B in Q1 FY26). The thesis is monetizing scarce 24/7 carbon-free + flexible capacity into multi-decade hyperscaler PPAs.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~94¢ is cost of goods and ~0¢ operating expense, leaving ~6¢ of operating profit (~9¢ net).
Revenue trend
Margins
up (quarter-specific; mix + spreads)
down (net income -38% YoY FY25 on prior-year one-offs)
up (Calpine accretion ~$2/sh)
up (2028–29 FCF guide $11.5–13.0B)
COGS structure
COGS is dominated by purchased power and fuel — natural gas for the Calpine fleet (Calpine fuel + purchased-power ~$6.35B in Q1 FY26; cost of energy/fuel rose ~14% on gas prices) plus nuclear fuel. Nuclear has very low marginal fuel cost (the high-margin baseload), so consolidated cost sensitivity now skews to natural-gas prices post-Calpine. Other costs: O&M on the nuclear/gas fleet, refueling-outage costs, and PJM capacity/transmission charges.
Capex
~$3.9B capital plan funds ~1 GW of nuclear uprates over the decade (e.g. ~135 MW at Braidwood/Byron for double-digit returns), the Crane/Three Mile Island Unit-1 restart (~835 MW, targeted ~2027–28), new gas (Freestone delivery targeted Q4 2026) + battery storage, and ~35% allocated to building nuclear-fuel inventory (fact).
Latest earnings
Beat — adjusted EPS $2.74 vs ~$2.59 consensus (~+6%); revenue $11.1B vs ~$9.0B expected; GAAP EPS $4.49 (fact). Stock +~4% pre-market on the print.
FY2026 adjusted EPS $11.00–$12.00 affirmed; FCF (before growth) $8.4B across 2026–27, rising to $11.5–13.0B in 2028–29; 20%+ base EPS CAGR framing + $5B buyback (fact)
- Adjusted EPS (Q1 FY26)
- $2.74 (beat $2.59)
- Revenue (Q1 FY26)
- $11.1B (+64% YoY)
- Nuclear capacity factor (FY25)
- 94.7% (ex-Salem/STP) — fact
- Calpine accretion (FY26)
- ~$2/share
Growth drivers
- Direct hyperscaler PPAs — Microsoft 835 MW (Crane/TMI restart, ~2028) and Meta 1,121 MW (Clinton, 2027) — 20-year contracts that de-risk merchant exposure (fact)
- Calpine integration — ~$2/sh accretion, +23 GW gas/geothermal, dispatchable/flexible capacity hyperscalers also need (fact)
- Nuclear uprates — ~1 GW of incremental carbon-free capacity at double-digit returns; ~5,000 MW submitted into PJM queue (uprates + gas + storage) (fact)
- Rising PJM/Northeast capacity prices + tightening reserve margins from data-center load growth (estimate)
- Capital return: $5B buyback authorization + ~10% annual dividend growth (fact)
- Policy tailwinds: nuclear PTC floor (IRA §45U) underpinning revenue, DOE support for restarts (fact)
Reported financials — SEC EDGAR
Audited GAAP figures pulled from SEC filings · latest filing 2026-02-24. The audited primary-source spine — not financial advice.
Revenue — annual (GAAP)
Margins & balance sheet — FY’25
Bull & bear
CEG is the single best-positioned vendor of the scarcest input in the AI buildout — 24/7 carbon-free baseload — and Calpine bolts on the flexible gas hyperscalers also need, all under a federal nuclear-PTC floor; the ~33% de-rate from the high hands you that franchise at ~13x EV/EBITDA with a $5B buyback and visible FCF growth to $11.5–13B by 2028–29.
- Irreplaceable asset: you cannot build a new ~21 GW nuclear fleet — the scarcity is structural, and AI load growth is a multi-year secular demand tailwind (fact/estimate)
- Contracted de-risking: Microsoft (835 MW) + Meta (1,121 MW) 20-year PPAs turn merchant power into utility-like contracted cash, with more of the 5,000 MW PJM queue to come (fact)
- Calpine accretion (~$2/sh) + 20%+ EPS-growth framing, FY26 adj-EPS guide $11–12 affirmed despite a tough Q1 op backdrop (fact)
- FCF inflection: $8.4B (2026–27) → $11.5–13.0B (2028–29) funds $5B buyback + ~10% dividend growth — capital return compounds the de-rated multiple (fact)
- PTC floor caps nuclear downside (IRA §45U), a structural risk-reducer competitors with no nuclear can't claim (fact)
The whole premium rests on hyperscaler PPA pricing staying scarce-and-rising; if AI capex digests or co-location gets regulated, CEG is a ~25x-P/E, ~$22B-net-debt, gas-price-exposed IPP whose 'growth' is mostly an acquisition — and the ~33% de-rate may be the market front-running exactly that.
- Demand-air-pocket risk: a pause in hyperscaler capex or a PPA-pricing reset removes the premium that justifies the multiple — the most likely cause of the de-rate (estimate)
- Leverage + gas exposure: ~$22B gross debt and a now gas-price-sensitive cost base (fuel +14% Q1 FY26) make the model more cyclical, not less (fact)
- Valuation: forward P/E ~25x vs IPP median ~17x and richer than Vistra (~10–12x EV/EBITDA) — priced for execution, little error margin (fact)
- Acquisition-flattered optics: +64% Q1 revenue is consolidation, not organic; integration of ~23 GW + ~$12.7B assumed debt carries real execution risk (fact)
- Regulatory tail: FERC/state pushback on behind-the-meter co-location could foreclose the highest-value deal structure (estimate)
What it is worth
Forward P/E + EV/EBITDA peer comps cross-checked with a reverse-DCF read on the implied growth
~$190–210
de-rate toward peer multiples (~17–18x) on AI-capex digestion / PPA-pricing reset / gas-cost squeeze, with leverage limiting buyback support
~$280–310
~24–26x mid-guide FY26 adj-EPS (~$11.5) with steady accretion + buyback; modest re-rate from the de-rated ~$268
~$360–400
re-rate back toward the high on multiple new behind-the-meter PPAs at premium CFE pricing + on-time restarts; ~30x on rising EPS
At ~$268 / ~25x FY26 adj-EPS (mid-guide ~$11.5) and ~13x EV/EBITDA, CEG trades at a clear premium to the IPP median (~17x P/E) and to Vistra (~10–12x EV/EBITDA) — the price implies sustained ~20%+ EPS growth AND durable hyperscaler PPA-pricing power; reasonable if AI power demand holds, rich if it digests.
SWOT
Strengths
- Largest US nuclear fleet (~94.7% FY25 capacity factor) — scarce, baseload, 24/7 carbon-free generation that is effectively irreplaceable on a 5–10yr horizon (fact)
- Calpine adds ~23 GW gas/geothermal → the only vendor selling hyperscalers BOTH 24/7 carbon-free AND flexible dispatchable power from one balance sheet (fact)
- Multi-decade hyperscaler PPAs (Microsoft, Meta) convert merchant volatility into contracted, inflation-linked cash flows (fact)
- Nuclear PTC floor (IRA §45U) puts a federal floor under nuclear revenue, capping downside (fact)
- Strong FCF generation guide ($8.4B 2026–27) funding $5B buyback + ~10% dividend growth (fact)
Weaknesses
- Post-Calpine leverage — ~$22B gross debt (LT ~$17.0B + ST ~$5.1B) vs ~$1.0B cash; ~$12.7B Calpine debt assumed — integration + deleveraging risk (fact)
- Now materially more exposed to natural-gas price swings via Calpine's gas fleet (Q1 FY26 fuel/purchased-power ~$6.35B; cost +14%) — fact
- Headline revenue growth is acquisition-driven, not organic — the +64% Q1 print flatters the underlying run-rate (fact)
- Long-dated, lumpy project execution (TMI/Crane restart, uprates) with regulatory + construction timing risk (estimate)
- Valuation premium — forward P/E ~25x vs IPP median ~17x leaves little error margin even after the de-rate (fact)
Opportunities
- Convert the ~5,000 MW PJM queue + uprate pipeline into additional hyperscaler PPAs at premium 24/7 CFE pricing (fact/estimate)
- Behind-the-meter / co-located data-center deals that bypass the grid-interconnection bottleneck (estimate)
- Further nuclear restarts / license extensions as AI power demand outpaces supply (estimate)
- Capacity-price upside as PJM/Northeast reserve margins tighten on data-center load (estimate)
- Geothermal optionality from the Calpine portfolio as a second 24/7 CFE product (estimate)
Threats
- Hyperscaler capex digestion / AI-demand-air-pocket would compress the PPA-pricing premium the whole thesis rests on (estimate — likely driver of the ~33% de-rate)
- Natural-gas price spikes squeezing Calpine-fleet spreads (fact-grounded sensitivity)
- Regulatory/FERC pushback on behind-the-meter co-location precedents that limit grid-bypass deals (estimate)
- Nuclear operational risk — an outage or safety event at a major unit hits the highest-margin output (estimate)
- Rate competition from peers (Vistra, Talen) chasing the same hyperscaler PPAs, plus SMR/new-build threats long-term (estimate)
Moats, dependencies & bottlenecks
Moats
Very high (10yr+) — no new large nuclear can be permitted/built on this horizon 94.7% FY25 capacity factor; the asset is effectively non-replicable (fact)
20-year contracts lock counterparties + pricing Converts merchant volatility to contracted cash; switching cost for hyperscalers is finding equivalent CFE elsewhere (fact)
Scale + dual-fuel breadth post-Calpine (~60 GW nuclear+gas+geothermal) Vistra/Talen are building the same combo Only vendor selling both 24/7 CFE AND flexible gas from one balance sheet today (fact)
Policy-dependent — durable while the statute stands Federal revenue floor competitors without nuclear can't access (fact)
Operating expertise / regulatory licenses (restart + uprate execution) embodied in licenses + talent, hard to replicate quickly TMI/Crane restart + uprate track record is a credibility moat for new PPAs (estimate)
Dependencies
Hyperscaler demand + willingness to pay premium PPA pricing (Microsoft, Meta, Amazon, Google) The entire growth thesis; an AI-capex digestion compresses the premium (estimate)
Calpine fuel/purchased-power ~$6.35B Q1 FY26; gas spikes squeeze gas-fleet spreads (fact)
~35% of capex to fuel-inventory build; Western enrichment capacity is thin post-Russia (fact/estimate)
NRC licensing, FERC co-location rulings) PTC floor is a tailwind; FERC behind-the-meter rulings could constrain the best deal structures (estimate)
Uncontracted volume still rides merchant spreads + capacity auctions (estimate)
Advantages
- Owns the scarcest AI input — the largest US nuclear baseload fleet, non-replicable on the investment horizon (fact)
- Only one-stop vendor of 24/7 carbon-free + flexible dispatchable gas after Calpine (fact)
- Contracted cash-flow visibility via 20-year hyperscaler PPAs + nuclear PTC floor (fact)
- Strong FCF + capital-return engine ($5B buyback, ~10% dividend growth) (fact)
- Proven restart/uprate execution credibility (TMI/Crane) that wins the next PPA (estimate)
Weaknesses
- High post-Calpine leverage (~$22B gross debt, ~$1.0B cash) and integration risk (fact)
- New, material natural-gas price sensitivity in the consolidated cost base (fact)
- Headline growth is acquisition-driven, masking modest organic growth (fact)
- Premium valuation (~25x fwd P/E) leaves little margin for execution/demand disappointment (fact)
- Lumpy, multi-year project timing (restarts, uprates) exposed to regulatory + construction delay (estimate)
- Concentration in PJM/Northeast + a handful of mega-counterparties raises tail-risk if any one PPA or market turns (estimate)
Bottlenecks
- Grid interconnection queues (PJM) — long, congested timelines gate how fast new uprates/gas/storage reach load; ~5,000 MW submitted but multi-year to clear (fact/estimate)
- Nuclear fuel inventory + enrichment supply — ~35% of capex going to fuel build signals a real supply constraint (fact)
- Nuclear regulatory/licensing throughput (NRC) for restarts (TMI/Crane) and uprates — sets the pace of new carbon-free MW (estimate)
- Skilled nuclear/construction labor for restart + uprate execution (estimate)
- Balance-sheet capacity / deleveraging post-Calpine limits how aggressively it can fund the next wave of growth capex (fact)
Top signals & trends
Top signals
Each new 20-year deal extends contracted cash + validates the premium (watch Amazon/Google/Oracle)
Favorable = unlocks the highest-value structure; restrictive = caps the upside (key swing factor)
On-time delivery de-risks the growth capex; slippage hits FCF guide (fact-anchored)
Drives the now-larger gas-fleet spread + uncontracted nuclear margin
Faster delever + synergy realization supports the ~$2/sh accretion and the buyback
Any datacenter-capex slowdown is the direct read-through to PPA pricing power
Trends
First sustained US electricity-demand growth in ~20 years; CEG sells the scarcest tranche (fact/estimate)
Drives premium pricing for nuclear specifically — CEG's core product (fact)
Near-term tailwind for incumbents like CEG; long-term SMR/new-build could erode scarcity (estimate)
Lifts merchant + capacity revenue on uncontracted volume (estimate)
Scale wins, but intensifies competition for the same hyperscaler PPAs (fact)
Helps nuclear spread economics but adds cost risk to the new Calpine gas fleet (fact)
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Nuclear fuel (uranium, conversion); CEG is building fuel inventory (~35% of capex)
US enrichment / HALEU supply — strategic for fuel security post-Russia
Coterra, Expand Energy) Fuel for the acquired Calpine gas fleet — now a material cost driver
GE Vernova / Westinghouse / turbine + reactor OEMs Turbines, reactor services, uprate engineering, future SMRs
Energy Capital Partners (former Calpine owner) Sold Calpine to CEG; ECP took CEG stock — now a large holder/partner
20-yr PPA, 835 MW from Crane/TMI Unit-1 restart (~2028) — anchor hyperscaler deal
20-yr PPA, 1,121 MW from Clinton Clean Energy Center (from 2027)
Major datacenter-power buyer in the same scarcity market; prospective PPA counterparty
24/7 CFE procurement leader; prospective hyperscaler PPA counterparty
Commercial & industrial + retail load (PJM/ERCOT/Northeast) Competitive retail supply book + wholesale market sales — the non-hyperscaler base
ERCOT, ISO-NE) Buyers of energy + capacity + ancillary services for uncontracted volume
~40+ GW gas/nuclear/solar/storage; bought Cogentrix gas (5.5 GW) + three 20yr Meta nuclear PPAs (2.6 GW) — closest dual-fuel rival, trades cheaper (~10–12x EV/EBITDA)
~15.6 GW incl 2.2 GW nuclear (Susquehanna); Amazon co-location pioneer + Western PJM acquisition; pure PJM nuclear+gas play chasing the same datacenter PPAs
~25 GW gas/coal + largest retail base (7–8M); 'bring-your-own-power' datacenter pivot but NO nuclear — can't sell 24/7 CFE baseload
Owns NJ nuclear (Salem/Hope Creek, partly co-owned with CEG); regulated-utility-plus-nuclear hybrid pursuing datacenter deals
Long-term threat — new nuclear (SMRs) + gas turbines could eventually erode the scarcity premium; also a supplier/partner today
Compete for datacenter load via regulated rate base; different model but absorb load CEG might otherwise serve